ASSESSMENT OF CONNECTIVITY DYNAMICS DURING NEURODEVELOPMENT THROUGH PERSISTENT HOMOLOGY IN A HIPSC-BIOPRINTED MODEL

Despite their importance in conditions like Autism Spectrum Disorder (ASD), the mechanisms driving circuit formation during neurodevelopment are still not fully understood. We investigated network connectivity in early development using 3D bioprinted neuronal constructs derived from human induced pluripotent stem cells (hiPSCs), a model offering superior physiological relevance compared to 2D cultures. Electrophysiological activity was recorded via a 120-channel microelectrode array (MEA) at 30, 60, 90 and 120 days in vitro (DIV). While standard multi-unit analysis showed increased firing rates over time, we employed persistent homology (PH), a topological data analysis method, to specifically investigate connectivity structure. After performing spike sorting (attached figure letter A, bottom panels), we generated smoothed spike count histograms (figure A, top panels), from which we constructed 3D phase space embeddings (figure B) to calculate simplicial complexes and obtain H1 invariants. While limited at DIV30 (figure C, 1st panel), H1 invariants birth-death distances effectively captured abrupt changes in firing bursts at DIV60 (figure C, 2nd panel). At DIV90 and DIV120 (figure C, 3rd and 4th panels respectively), these invariants displayed greater persistence during earlier cycles of simplicial detection. Goodness-of-fit testing on birth-death distance distributions confirmed that all groups originated from significantly distinct patterns (figure D, p < 0.001). These results indicate that H1 invariants can capture stable features embedded in both nuanced and abrupt firing transitions throughout maturation, highlighting the value of topological invariants for the future analysis of electrophysiological markers in developing neural tissue.